MCP Server Builder
anthropics/skills
Guides the design and implementation of Model Context Protocol servers in TypeScript or Python, from tool naming and error messages to evaluation.
Control Windows desktop applications: open/operate apps, click controls, type text, scroll, keyboard shortcuts, window management, toggle options, lists and combos, etc.
MITAuto-check passed
Install Windows Control
$ npx skills add different-ai-studio/teamclu --skill windows-control -a claude-codeProject install by default; add -g for ~/.claude/skills/.
$ gh skill install different-ai-studio/teamclu windows-control --agent claude-codeProject scope by default; add --scope user for a personal install. Needs GitHub CLI 2.90.0 or later (public preview).
$ git clone --depth 1 https://github.com/different-ai-studio/teamclu.git skills-src && mkdir -p .claude/skills && cp -r skills-src/packages/app/src/lib/skills/windows-control .claude/skills/windows-control && rm -rf skills-srcUse ~/.claude/skills/ instead of .claude/skills for a personal install. The folder must contain SKILL.md.
Claude Code skills documentation · loads skills from .claude/skills/
Install the "windows-control" agent skill from https://github.com/different-ai-studio/teamclu/tree/main/packages/app/src/lib/skills/windows-control into .claude/skills/windows-control/ in this project. Copy the whole folder (SKILL.md and every file beside it), keep the folder name "windows-control", then confirm the skill loads.Claude Code copies the folder itself, the same result as the manual copy. Check what it changed before you commit it.
$skill-installer install https://github.com/different-ai-studio/teamclu/tree/main/packages/app/src/lib/skills/windows-controlType this inside Codex. $skill-installer <name> installs a curated skill from openai/skills. The installer writes to $CODEX_HOME/skills (default ~/.codex/skills). Restart Codex if the skill does not show up.
$ npx skills add different-ai-studio/teamclu --skill windows-control -a codexProject install goes to .agents/skills/; add -g for ~/.codex/skills/.
$ gh skill install different-ai-studio/teamclu windows-control --agent codexProject scope by default (.agents/skills/); add --scope user for a personal install.
$ git clone --depth 1 https://github.com/different-ai-studio/teamclu.git skills-src && mkdir -p .agents/skills && cp -r skills-src/packages/app/src/lib/skills/windows-control .agents/skills/windows-control && rm -rf skills-srcUse ~/.agents/skills/ instead of .agents/skills for a personal install.
Codex skills documentation · loads skills from .agents/skills/
Install the "windows-control" agent skill from https://github.com/different-ai-studio/teamclu/tree/main/packages/app/src/lib/skills/windows-control into .agents/skills/windows-control/ in this project. Copy the whole folder (SKILL.md and every file beside it), keep the folder name "windows-control", then confirm the skill loads.Codex copies the folder itself, the same result as the manual copy. Check what it changed before you commit it.
$ npx skills add different-ai-studio/teamclu --skill windows-control -a cursorProject install goes to .agents/skills/; add -g for ~/.cursor/skills/.
$ gh skill install different-ai-studio/teamclu windows-control --agent cursorProject scope by default (.agents/skills/); add --scope user for a personal install.
$ git clone --depth 1 https://github.com/different-ai-studio/teamclu.git skills-src && mkdir -p .cursor/skills && cp -r skills-src/packages/app/src/lib/skills/windows-control .cursor/skills/windows-control && rm -rf skills-srcUse ~/.cursor/skills/ instead of .cursor/skills for a personal install.
Cursor skills documentation · loads skills from .cursor/skills/, .agents/skills/, .claude/skills/, .codex/skills/
Install the "windows-control" agent skill from https://github.com/different-ai-studio/teamclu/tree/main/packages/app/src/lib/skills/windows-control into .cursor/skills/windows-control/ in this project. Copy the whole folder (SKILL.md and every file beside it), keep the folder name "windows-control", then confirm the skill loads.Cursor copies the folder itself, the same result as the manual copy. Check what it changed before you commit it.
$ gemini skills install https://github.com/different-ai-studio/teamclu.git --path packages/app/src/lib/skills/windows-control--scope user (default) or --scope workspace; --path is the subfolder of the repo that holds the skill; --consent skips the security confirmation prompt.
$ npx skills add different-ai-studio/teamclu --skill windows-control -a gemini-cliProject install goes to .agents/skills/; add -g for ~/.gemini/skills/.
$ gh skill install different-ai-studio/teamclu windows-control --agent gemini-cliProject scope by default (.agents/skills/); add --scope user for a personal install.
$ git clone --depth 1 https://github.com/different-ai-studio/teamclu.git skills-src && mkdir -p .gemini/skills && cp -r skills-src/packages/app/src/lib/skills/windows-control .gemini/skills/windows-control && rm -rf skills-srcUse ~/.gemini/skills/ instead of .gemini/skills for a personal install, then run /skills reload.
Gemini CLI skills documentation · loads skills from .gemini/skills/, .agents/skills/
Install the "windows-control" agent skill from https://github.com/different-ai-studio/teamclu/tree/main/packages/app/src/lib/skills/windows-control into .gemini/skills/windows-control/ in this project. Copy the whole folder (SKILL.md and every file beside it), keep the folder name "windows-control", then confirm the skill loads.Gemini CLI copies the folder itself, the same result as the manual copy. Check what it changed before you commit it.
$ gh skill install different-ai-studio/teamclu windows-controlInstalls for Copilot at project scope by default; add --scope user for a personal install. Preview a skill first with gh skill preview. Needs GitHub CLI 2.90.0 or later (public preview).
$ npx skills add different-ai-studio/teamclu --skill windows-control -a github-copilotProject install goes to .agents/skills/; add -g for ~/.copilot/skills/.
$ git clone --depth 1 https://github.com/different-ai-studio/teamclu.git skills-src && mkdir -p .github/skills && cp -r skills-src/packages/app/src/lib/skills/windows-control .github/skills/windows-control && rm -rf skills-srcUse ~/.copilot/skills/ instead of .github/skills for a personal install. Commit .github/skills so cloud agent and code review can use it.
GitHub Copilot skills documentation · loads skills from .github/skills/, .claude/skills/, .agents/skills/
Install the "windows-control" agent skill from https://github.com/different-ai-studio/teamclu/tree/main/packages/app/src/lib/skills/windows-control into .github/skills/windows-control/ in this project. Copy the whole folder (SKILL.md and every file beside it), keep the folder name "windows-control", then confirm the skill loads.GitHub Copilot copies the folder itself, the same result as the manual copy. Check what it changed before you commit it.
$ npx skills add different-ai-studio/teamclu --skill windows-control -a opencodeOpenCode documents no install command of its own. Project install goes to .agents/skills/; add -g for ~/.config/opencode/skills/.
$ gh skill install different-ai-studio/teamclu windows-control --agent opencodeProject scope by default (.agents/skills/); add --scope user for a personal install.
$ git clone --depth 1 https://github.com/different-ai-studio/teamclu.git skills-src && mkdir -p .opencode/skills && cp -r skills-src/packages/app/src/lib/skills/windows-control .opencode/skills/windows-control && rm -rf skills-srcUse ~/.config/opencode/skills/ instead of .opencode/skills for a personal install.
OpenCode skills documentation · loads skills from .opencode/skills/, .claude/skills/, .agents/skills/
Install the "windows-control" agent skill from https://github.com/different-ai-studio/teamclu/tree/main/packages/app/src/lib/skills/windows-control into .opencode/skills/windows-control/ in this project. Copy the whole folder (SKILL.md and every file beside it), keep the folder name "windows-control", then confirm the skill loads.OpenCode copies the folder itself, the same result as the manual copy. Check what it changed before you commit it.
Facts
- Skill name
windows-control- GitHub stars
- 167
- Token cost
- ~1.7k tokens
- SKILL.md length
- 560 words
- Files
- 1
- Skills in repo
- 8
- Repo updated
- First seen
- Licence
- MIT
At a glance
Control Windows desktop applications: open/operate apps, click controls, type text, scroll, keyboard shortcuts, window management, toggle options, lists and combos, etc.
- Works in 3 steps: Probe (Inspect UIA / window) —… → Route (Decision Routing) — Choose Layer… → Act (Execute Operation) — Perform the…
- Words: open app
- SKILL.md covers Decision: Unified Flow (Probe…, Layer 1: PowerShell + Shell, Layer 2: autoui-mcp-server… and Guardrails
- Instructions only: no scripts, shell commands, URLs or credentials in SKILL.md
What it does
Windows Control is an agent skill from different-ai-studio/teamclu. Control Windows desktop applications: open/operate apps, click controls, type text, scroll, keyboard shortcuts, window management, toggle options, lists and combos, etc. Note: NEVER invoke this skill for opening web pages/visiting websites — use browser-related tools instead. Trigger words: open app, operate app, control computer, click, type, scroll, switch, dropdown, select, modify, change to, set, check, uncheck, app-control, toggle.
Its SKILL.md is about 1.7k tokens, which your agent loads only when the skill is triggered. It is a single SKILL.md file with no bundled scripts. Compatibility notes: Windows only. Layer 2 requires autoui-mcp-server MCP server.
It works with Model Context Protocol. The repository describes itself as: TeamClu, AI Agent Desktop Workspace. The licence is MIT.
When your agent uses it
- Words: open app
- Control computer
Example prompts
- “/windows-control”
Requirements
- Compatibility (from SKILL.md): Windows only. Layer 2 requires autoui-mcp-server MCP server.
Workflow steps
3 steps, taken from the first numbered list in SKILL.md.
- Probe (Inspect UIA / window) — Lightweight read-only checks that the target window/process exists and automation peers are discoverable.
- Route (Decision Routing) — Choose Layer 1 or Layer 2 from probe results.
- Act (Execute Operation) — Perform the operation in the selected layer.
What it can do on your machine
Read from SKILL.md and the folder at commit 0ea73c3. It shows what the files ask for, not the result of running them.
Tool permissions
Pre-approves nothing: there is no allowed-tools line, so your agent's usual permission prompts apply.
From allowed-tools in the SKILL.md frontmatter.
Runs code
No scripts in the folder and no shell commands in SKILL.md (its code samples are powershell).
From the folder's file list and the shell code blocks in SKILL.md.
Network
No URLs in SKILL.md.
From URLs in SKILL.md, links to its own repository left out.
Credentials
Names no API keys, tokens, secrets or passwords.
From names ending in _API_KEY, _TOKEN, _SECRET, _KEY or _PASSWORD in SKILL.md.
Compatibility
Windows only. Layer 2 requires autoui-mcp-server MCP server.
From compatibility in the SKILL.md frontmatter.
Context cost
Windows Control loads about 1.7k tokens when it runs. Until then it costs about 114 tokens; SKILL.md has 560 words of instructions outside code blocks.
Estimates: characters ÷ 4, the usual rule of thumb; real counts depend on the model's tokenizer. Scripts and assets cost tokens only if the agent reads them.
Safety
Auto-check passed
The automated check found no risky patterns in SKILL.md.
Automated static check — not a guarantee. Review scripts before installing. It scans the text of SKILL.md for risky patterns (piping downloads into a shell, reading credential files, hidden Unicode, destructive commands); files beside SKILL.md are not scanned.
SKILL.md
The full file from different-ai-studio/teamclu at commit 0ea73c3, republished under its MIT licence (© different-ai-studio). 560 words, ~1,687 tokens.
Download SKILL.mdSave it as
.claude/skills/windows-control/SKILL.md (or your agent's skills folder).- name
- windows-control
- description
- Control Windows desktop applications: open/operate apps, click controls, type text, scroll, keyboard shortcuts, window management, toggle options, lists and combos, etc. Note: NEVER invoke this skill for opening web pages/visiting websites — use browser-related tools instead. Trigger words: open app, operate app, control computer, click, type, scroll, switch, dropdown, select, modify, change to, set, check, uncheck, app-control, toggle.
- compatibility
- Windows only. Layer 2 requires autoui-mcp-server MCP server.
Windows Control
Two-layer strategy: Probe first, route fast. For all apps, probe UI Automation (UIA) accessibility first to confirm the tree is usable and contains the target element, then use Layer 1; otherwise switch to Layer 2 immediately to avoid misoperations on an incomplete tree.
Decision: Unified Flow (Probe → Route → Act)
All apps follow the same flow:
- Probe (Inspect UIA / window) — Lightweight read-only checks that the target window/process exists and automation peers are discoverable.
- Route (Decision Routing) — Choose Layer 1 or Layer 2 from probe results.
- Act (Execute Operation) — Perform the operation in the selected layer.
Probe Step (Mandatory, Before Any Operation)
Probe in read-only mode — no clicks, no focus changes that perform actions:
# 1. Process exists
Get-Process -Name "Notepad" -ErrorAction SilentlyContinue
# 2. List top-level windows with title (filter empty titles if needed)
Get-Process | Where-Object { $_.MainWindowTitle } | Select-Object Name, Id, MainWindowTitleFor deeper UIA inspection from PowerShell (when .NET UIA is available):
Add-Type -AssemblyName UIAutomationClient
Add-Type -AssemblyName UIAutomationTypes
$root = [Windows.Automation.AutomationElement]::RootElement
# Walk children or use WindowPattern — scope narrowly to the target process/main windowIf UIA APIs are unavailable or the tree is empty/generic, treat as fail probe → Layer 2.
Route Decision Rules
| Probe Result | Route | Reason |
|---|---|---|
| Rich control tree; target element identifiable | → Layer 1 | UIA usable |
| Tree present but target not found | → Layer 2 | Wrong or partial tree |
Tree sparse (few elements) or mostly Pane/Custom with no names | → Layer 2 | Unreliable |
| Errors (access denied, process not found, timeout) | → Layer 2 (or ask user to run as appropriate / grant access) | UIA unavailable |
| Many unnamed duplicates | → Layer 2 | Ambiguous |
If in doubt after Probe, choose Layer 2.
Out of scope: Web page operations → browser tools only.
Reverse constraint: This skill targets local Windows desktop applications. Never use Playwright MCP
browser_*tools (browser_click,browser_select_option,browser_fill_form,browser_type,browser_press_key,browser_hover,browser_drag,browser_snapshot,browser_navigate,browser_screenshot, …) for desktop UI — they only affect web pages. For desktop screenshots or vision steps, use autoui-mcp-server (auto_vision_*,auto_mouse_*,auto_keyboard_*).
Layer 1: PowerShell + Shell
Run powershell.exe, pwsh.exe, cmd, start, and Windows utilities via the Shell tool.
Prerequisite: Probe indicates UIA (or window focus strategy) is viable. Otherwise skip to Layer 2.
Resolve Application
# Installed apps (Start Menu shortcuts — adjust user name if needed)
Get-ChildItem "$env:ProgramData\Microsoft\Windows\Start Menu\Programs" -Recurse -Filter *.lnk -ErrorAction SilentlyContinue | Select-Object FullName
Get-ChildItem "$env:APPDATA\Microsoft\Windows\Start Menu\Programs" -Recurse -Filter *.lnk -ErrorAction SilentlyContinue | Select-Object FullName
# Running processes with UI
Get-Process | Where-Object MainWindowHandle -ne 0 | Select-Object Name, Id, MainWindowTitleOpen / Focus / Close
Before opening: Resolve the exact executable or shortcut name. Prefer
Start-Processwith-WindowStyle Normal. Avoid forcing fullscreen.
# Launch
Start-Process "notepad"
Start-Process "C:\Path\App.exe" -ArgumentList '--flag'
# Bring existing to foreground (conceptual — may require Win32 ShowWindow in some cases)
# Prefer UIA / Layer 2 if focus APIs misbehave.
# Close gracefully
Stop-Process -Name "Notepad" -ErrorAction SilentlyContinueFile / URI
Start-Process "C:\path\file.txt" # Default handler
Start-Process "msedge" "https://example.com" # Example browser — still, prefer browser tools for *web* tasksShow full SKILL.md (232 more words)Show less
UIA-Oriented Operations (When Probe Passed)
Use narrowly scoped scripts: find the automation element by AutomationId, Name, or control type, then invoke InvokePattern, ValuePattern, TogglePattern, etc. If scripting becomes brittle, switch to Layer 2.
Keyboard Shortcuts (Windows)
- Ctrl replaces macOS Cmd for most shortcuts (e.g. save: Ctrl+S).
- Alt accesses menus; Alt+F4 closes foreground window.
- Win opens Start; Win+D shows desktop.
Layer 2: autoui-mcp-server Vision Automation
Same role as on macOS: when UIA is insufficient or visual verification is required.
Do not skip Probe to jump here without reason; once Probe fails, use Layer 2 promptly.
Workflow: plan → act → verify
auto_vision_plan(intent='...', context=[...])auto_mouse_click/auto_keyboard_type/auto_mouse_scroll/auto_mouse_dragas appropriateauto_vision_verify(assertion='...')
Notes (Windows)
- Double-click to open files in Explorer may need
clicks=2. - Focus before scroll: click the target region first.
- Never guess coordinates — re-plan when the layout changes.
Tool List
| Tool | Purpose |
|---|---|
auto_vision_plan | Plan + candidate elements |
auto_vision_verify | Visual assertion |
auto_mouse_click | Click (incl. double) |
auto_mouse_move | Hover |
auto_mouse_scroll | Scroll |
auto_mouse_drag | Drag |
auto_keyboard_type | Text entry |
auto_keyboard_press | Key chords |
Guardrails
- Never run destructive commands (
Format-Volume, massRemove-Item -Recurseon system paths, etc.) without explicit user confirmation. - Never weaken security products or UAC prompts automatically.
- Never type secrets/passwords unless the user supplied them explicitly for that step.
- Probe is read-only — no mutating UI actions during Probe.
- Prefer Layer 2 when UIA is ambiguous; verify after Layer 2 steps.
- No Playwright
browser_*for desktop UI; use autouiauto_*only.
© different-ai-studio, MIT. Rendered from Markdown: HTML in the file is shown as text, images as links, and headings moved down two levels. Raw file
Files
Just SKILL.md in packages/app/src/lib/skills/windows-control of different-ai-studio/teamclu.
Open the folder on GitHubat commit 0ea73c3
Compare with similar skills
Windows Control next to the 5 skills that share the most tags, products or categories with it. Stars are the repository's; “used in” counts other GitHub owners with a copy.
| Skill | Stars | Used in | Tokens | Auto-check | Licence | Repo updated |
|---|---|---|---|---|---|---|
| Windows Control this skilldifferent-ai-studio/teamclu | 167 | — | ~1.7k | Automated safety check: Pass | MIT | |
| MCP Server Builderanthropics/skills | 180k | 62 repos | ~2.3k | Automated safety check: Pass | Apache-2.0 | |
| MCP Server BuildershareAI-lab/learn-claude-code | 78k | 5 repos | ~1.2k | Automated safety check: Pass | MIT | |
| MCP Integration for Pluginsanthropics/claude-plugins-official | 37k | 11 repos | ~3.1k | Automated safety check: Pass | Apache-2.0 | |
| Figma use_figma Plugin API Ruleswarpdotdev/warp | 65k | 4 repos | ~4.4k | Automated safety check: Pass | AGPL-3.0 | |
| Stitch to Remotion Walkthrough Videosgoogle-labs-code/stitch-skills | 8.4k | 6 repos | ~3.2k | Automated safety check: Notes | Apache-2.0 |
Similar skills
- Official
MCP Server Builder
shareAI-lab/learn-claude-code
Walks through building MCP servers in Python or TypeScript that expose tools, resources and prompts to Claude, with templates, registration and testing.
- Official
MCP Integration for Plugins
anthropics/claude-plugins-official
Explains how to bundle Model Context Protocol servers in a Claude Code plugin, covering config files, stdio, SSE, HTTP and WebSocket server types, and authentication.
Figma use_figma Plugin API Rules
warpdotdev/warp
Required groundwork before any use_figma call: the rules and reference files for running JavaScript in a Figma file through the Plugin API without common failures.
- Official
Stitch to Remotion Walkthrough Videos
google-labs-code/stitch-skills
Builds walkthrough videos from Stitch design projects using Remotion, with transitions, zoom effects and text overlays on each screen.
MCP Development
coollabsio/coolify
A skill your agent uses for Laravel MCP development. An agent skill from coollabsio/coolify.
More from different-ai-studio/teamclu
All 8 skills in this repoApp Auth
different-ai-studio/teamclu
A skill your agent uses when implementing TeamClu platform sign-in, organization-role access, employee pages or protected data endpoints in an app, or changing its login permissions.
Create Role
different-ai-studio/teamclu
Create a workspace role that follows the ROLE.md specification.
Image Gen
different-ai-studio/teamclu
生成图片：配图、插图、海报、头像、示意图、Logo 草稿、封面、banner。触发词：画一张、画个、生成图片、生成一张图、做张图、做个图、配图、出图、生图、AI 绘图、image、draw、illustration、poster。仅从零生成，不做已有图片的编辑或修改。
macOS Control
different-ai-studio/teamclu
Control macOS desktop applications: open/operate apps, click buttons, type text, scroll, keyboard shortcuts, window management, toggle options, dropdown selections, etc.
Sentry Fix
different-ai-studio/teamclu
A skill your agent uses when the user wants to fix a Sentry issue, auto-repair a bug from Sentry, or create a fix PR for a Sentry error.
Sentry Monitor
different-ai-studio/teamclu
A skill your agent uses when the user wants to check Sentry issues, run a Sentry daily report, or monitor error trends.
Related
Works with
Questions about Windows Control
What does Windows Control do?
Control Windows desktop applications: open/operate apps, click controls, type text, scroll, keyboard shortcuts, window management, toggle options, lists and combos, etc. Windows Control is an agent skill from different-ai-studio/teamclu. Control Windows desktop applications: open/operate apps, click controls, type text, scroll, keyboard shortcuts, window management, toggle options, lists and combos, etc.
When should I use Windows Control?
Windows Control fits situations like: words: open app; control computer.
How do I install Windows Control in Claude Code?
Run `npx skills add different-ai-studio/teamclu --skill windows-control -a claude-code`. Or copy the skill folder (packages/app/src/lib/skills/windows-control in different-ai-studio/teamclu) into .claude/skills/windows-control in your project. Claude Code loads it when a task matches its description.
How do I install Windows Control in Codex?
Run `npx skills add different-ai-studio/teamclu --skill windows-control -a codex`. Or copy the skill folder (packages/app/src/lib/skills/windows-control in different-ai-studio/teamclu) into .agents/skills/windows-control in your project. Codex loads it when a task matches its description.
Can I use Windows Control in Cursor, Gemini CLI or GitHub Copilot?
Cursor, Gemini CLI, GitHub Copilot and OpenCode also load SKILL.md folders. With the skills CLI, run `npx skills add different-ai-studio/teamclu --skill windows-control -a cursor` (or -a gemini-cli, github-copilot or opencode for the others). To copy it by hand, put the folder in .cursor/skills/windows-control, .gemini/skills/windows-control, .github/skills/windows-control and .opencode/skills/windows-control in your project.
What does Windows Control need to run?
SKILL.md names no scripts, command-line tools or credentials: Windows Control is instructions for the agent only. Compatibility (from SKILL.md): Windows only. Layer 2 requires autoui-mcp-server MCP server..
Does Windows Control access the network?
SKILL.md contains no URLs. Any network use would come from the scripts or tools the agent runs. This is read from the text; nothing was executed.
Is Windows Control safe to install?
Our automated static check of SKILL.md found no risky patterns, such as piping downloads into a shell, reading credential files or hidden Unicode. It is not a guarantee. Review the folder before installing.
What licence does Windows Control use?
Windows Control is published under the MIT licence (the repository's licence). It allows redistribution, so the full SKILL.md is shown on this page.
How many tokens does Windows Control use?
About 1.7k tokens (SKILL.md is roughly 6.7k characters). Agents keep only the skill's name and description in context until a task matches; then they load SKILL.md in full.
What are the alternatives to Windows Control?
Skills that share tags, products or a category with Windows Control: MCP Server Builder (anthropics/skills, 180k stars), MCP Server Builder (shareAI-lab/learn-claude-code, 78k stars), MCP Integration for Plugins (anthropics/claude-plugins-official, 37k stars) and Figma use_figma Plugin API Rules (warpdotdev/warp, 65k stars). The comparison table on this page puts their stars, adoption, token cost, safety result and licence side by side.
Who maintains Windows Control?
different-ai-studio (a GitHub organization) maintains it in different-ai-studio/teamclu, which has 167 GitHub stars. The repository holds 8 skills in this directory. The repository was last updated on October 6, 2026.
Source: different-ai-studio/teamclu on GitHub. Facts on this page come from the repository at the commit we read; the author's words are quoted as theirs.